INDEX

Explanations

terms related to locations or places

oai_token-act-pair · gpt-3.5-turbo

mentions of location-related terms

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

tul

-0.63

toe

-0.61

fo

-0.60

 posted

-0.60

 screw

-0.58

Dark

-0.57

te

-0.57

 waist

-0.56

 Nich

-0.55

 railing

-0.55

POSITIVE LOGITS

ocation

4.40

ocations

3.47

ocated

3.17

ocating

2.95

ocate

2.66

ocative

2.11

oc

1.30

oca

1.26

oking

1.23

ocal

1.21

Activations Density 0.017%

terms related to locations or places

mentions of location-related terms

No Comments

No Known Activations